Confidential AI Network

Governed multi-party training without a central data lake

High-value models often need data that cannot be freely centralized. Confidential AI Network (CAN) explores a control model where providers, consumers, and clean-room operators negotiate a machine-enforceable contract, train only in policy-bound environments, and retain provenance for verification.

Executive overview (CISO) Ricardian contracts CAN: contract to prediction Product tour

Problem

Healthcare, finance, public sector, and industrial organizations hold data that cannot be freely exported. Model builders need that data, but a shared central lake concentrates sovereignty, liability, and competitive risk.

CAN treats collaboration as a protocol: catalog metadata → Ricardian contract → signed, policy-bound training → provenance on a confidential ledger. Informed by India’s iSPIRT DEPA (Data Empowerment and Protection Architecture); intended to run on enterprise clouds with native IdPs.

Design objectives

Architecture at a glance

One control surface, three trust planes: people, cloud APIs, and peer workloads. Contracts bind the commercial agreement; confidential compute and cryptography enforce the technical one.

Shared control objectives across Azure, AWS, GCP, and OCI: Multi-cloud security architecture patterns.

  ┌─────────────┐     ┌─────────────┐     ┌─────────────┐
  │     TDP     │     │     TDC     │     │  TSP/CCRP   │
  │  datasets   │     │   models    │     │ clean rooms │
  └──────┬──────┘     └──────┬──────┘     └──────┬──────┘
         │                   │                   │
         └─────────┬─────────┴─────────┬─────────┘
                   ▼                   ▼
         ┌─────────────────────────────────────┐
         │  Portal + APIs  ·  Ricardian terms  │
         │  Cloud identity provider SSO · roles│
         └──────────────────┬──────────────────┘
                            ▼
         ┌─────────────────────────────────────┐
         │  Contract lifecycle → training jobs │
         │  DEK/MEK · DP options · provenance  │
         └──────────┬───────────────┬──────────┘
                    ▼               ▼
         ┌────────────────┐  ┌─────────────────┐
         │ Isolated train │  │ SCITT CCF ledger│
         │ TEE / K8s / VM │  │ claims & audit  │
         └────────────────┘  └─────────────────┘

Control plane

Portal, APIs, contract state machine, catalog, AppAdmin; cloud IdP for humans.

Data & crypto plane

Dataset encryption, key escrow, optional DP, DEPA-aligned entity IDs.

Execution plane

Policy-bound training; SPIFFE/SPIRE for east-west where enabled.

Evidence plane

SCITT CCF claims; SIEM export for SOC workflows.

Start here

Product tour (Local path) · Azure confidential computing (threat model · Key Vault · SKR).

Notes & whitepapers

Long-form specs live in the repository docs/ tree; this site is the short path.

Start here

Executive overview

Executive path — start here before specialist notes.

  1. Contract management in CAN — party signing keys, sign gates, and verification

    August 17, 2026 · UserKey · authz vs crypto verify · Key Vault target

  2. Ricardian contracts in CAN — legal prose the runtime can enforce

    August 16, 2026 · Dual-layer legal + machine binding · sign → SIGNED → train

  3. Azure product tour deck — Entra to governed prediction

    August 16, 2026 · Stakeholder slide deck for the Azure deployment narrative

  4. Confidential AI Network: from signed contract to governed prediction

    August 14, 2026 · Train → infer under Open-GMASE, with CompliancePulse ingest

  5. Governed AI for the enterprise — a CISO’s overview

    August 14, 2026 · Executive brief on CAN, Open-GMASE, and CompliancePulse

  6. Product tour

    Screenshots · CAN end-to-end path

  7. Try it: CAN ↔ Open-GMASE ↔ CompliancePulse (demo slice)

    August 14, 2026 · Live OPA → AuditLogs → CP ingest seam (research demo)

Specialist notes

Platform — Confidential AI Network

Product and architecture detail for CAN.

  1. Contract management in CAN — party signing keys, sign gates, and verification

    August 17, 2026 · What signs the agreement vs what unlocks DEK/MEK

  2. Ricardian contracts in CAN — legal prose the runtime can enforce

    August 16, 2026 · What the agreement is and how train/infer/audit hang off it

  3. Confidential AI Network: from signed contract to governed prediction

    August 14, 2026 · Current product surface — contract through governed inference

  4. Building Confidential AI Network — governed multi-party training without a data lake

    July 29, 2026 · Technical product walkthrough

KMS, TEE & provenance

Keys, clean-room decrypt gates, and tamper-evident audit when models misbehave.

  1. Azure confidential computing for CAN — threat model, Key Vault, SKR, e2e training

    August 17, 2026 · Attestation · Secure Key Release · dual-key escrow on Azure

  2. KMS for Confidential AI Network — DEK, MEK, and dual-key escrow

    August 16, 2026 · Principal-owned keys · escrow · cloud KMS

  3. TEE training in CAN — attest, verify contract, then decrypt

    August 16, 2026 · Hardware attestation · decrypt-in-memory · live vs target

  4. Merkle trees for model audit — when the model misbehaves

    August 16, 2026 · Tamper-evident provenance · inclusion proofs · incident playbook

Agent governance — G-MASE & CompliancePulse

For security architects and platform engineers.

  1. G-MASE deep dive: Governed Multi-Agent SecOps Environment

    August 14, 2026 · Swarm topology, control order, Open-GMASE Core

  2. CompliancePulse AI deep dive: zero-trust control plane

    August 14, 2026 · Open-core product path, ingest, roadmap

  3. Governing autonomous AI agents in cybersecurity operations

    July 31, 2026 · Attack matrix & multi-cloud IAM whitepaper

  4. Unified Governed Agentic SecOps Framework

    August 14, 2026 · Swarm + control plane together

  5. Try it: CAN ↔ Open-GMASE ↔ CompliancePulse (demo slice)

    August 14, 2026 · Runnable seam for stakeholders (incl. CP ingest)

Architecture essays

Adjacent notes on protocols and agentic systems (not CAN product docs).

  1. Is India’s OCEN protocol the missing blueprint for agentic AI workflows?

    August 23, 2026 · Andrew Ng workflow redesign · OCEN / DPI as multi-party rails · LinkedIn

Identity — humans, cloud APIs, workloads

Zero-trust identity for people and for agent fleets on Kubernetes.

  1. Three identity planes: humans, cloud APIs, and workloads

    July 28, 2026 · Conceptual model

  2. SPIFFE/SPIRE with Azure — AKS Workload Identity and Entra federation

    August 17, 2026 · Path N / Path F · exact SPIFFE subjects

  3. SPIFFE/SPIRE with OCI IAM workload identity

    July 28, 2026 · CAN scaffolding

  4. The credential every AI agent fleet on an OKE node ends up sharing

    July 29, 2026 · Full write-up

  5. Beyond instance principals: fixing the pod identity gap in OKE

    July 29, 2026 · Deep dive

  6. Short take: per-node credentials break multi-agent fleets on OKE

    July 29, 2026 · LinkedIn-length summary

Cloud security architectures

Deployment-shaped notes for cloud identity and edge controls.

  1. SPIFFE/SPIRE with Azure — AKS Workload Identity and Entra federation

    August 17, 2026 · Design · AZURE_SPIFFE_SPIRE_WIF.md

  2. Azure confidential computing for CAN — threat model, Key Vault, SKR, e2e training

    August 17, 2026 · Attestation · Secure Key Release · dual-key train path

  3. Azure product tour deck — Entra to governed prediction

    August 16, 2026 · E2E stakeholder deck · open slides

  4. Azure security architecture — Entra-only identity on cloud

    July 28, 2026 · Azure / Entra

Compliance & documentation map

Where reviewers find controls evidence and how requirements map to NIST, CIS, and OWASP (incl. LLM).

  1. Where to find CAN security docs (map for reviewers)

    July 28, 2026 · Docs index

  2. Requirements met — NIST, CIS & OWASP (incl. LLM) mapping

    July 28, 2026 · GRC mapping

All posts by date

Newest first.